Showing papers by "Dalian University of Technology published in 2018"

Deep Mutual Learning

Abstract: Model distillation is an effective and widely used technique to transfer knowledge from a teacher to a student network The typical application is to transfer from a powerful large network or ensemble to a small network, in order to meet the low-memory or fast execution requirements In this paper, we present a deep mutual learning (DML) strategy Different from the one-way transfer between a static pre-defined teacher and a student in model distillation, with DML, an ensemble of students learn collaboratively and teach each other throughout the training process Our experiments show that a variety of network architectures benefit from mutual learning and achieve compelling results on both category and instance recognition tasks Surprisingly, it is revealed that no prior powerful teacher network is necessary - mutual learning of a collection of simple student networks works, and moreover outperforms distillation from a more powerful yet static teacher

Progressive Attention Guided Recurrent Network for Salient Object Detection

Abstract: Effective convolutional features play an important role in saliency estimation but how to learn powerful features for saliency is still a challenging task. FCN-based methods directly apply multi-level convolutional features without distinction, which leads to sub-optimal results due to the distraction from redundant details. In this paper, we propose a novel attention guided network which selectively integrates multi-level contextual information in a progressive manner. Attentive features generated by our network can alleviate distraction of background thus achieve better performance. On the other hand, it is observed that most of existing algorithms conduct salient object detection by exploiting side-output features of the backbone feature extraction network. However, shallower layers of backbone network lack the ability to obtain global semantic information, which limits the effective feature learning. To address the problem, we introduce multi-path recurrent feedback to enhance our proposed progressive attention driven framework. Through multi-path recurrent connections, global semantic information from the top convolutional layer is transferred to shallower layers, which intrinsically refines the entire network. Experimental results on six benchmark datasets demonstrate that our algorithm performs favorably against the state-of-the-art approaches.

Atomic-level insight into super-efficient electrocatalytic oxygen evolution on iron and vanadium co-doped nickel (oxy)hydroxide.

Abstract: It is of great importance to understand the origin of high oxygen-evolving activity of state-of-the-art multimetal oxides/(oxy)hydroxides at atomic level. Herein we report an evident improvement of oxygen evolution reaction activity via incorporating iron and vanadium into nickel hydroxide lattices. X-ray photoelectron/absorption spectroscopies reveal the synergistic interaction between iron/vanadium dopants and nickel in the host matrix, which subtly modulates local coordination environments and electronic structures of the iron/vanadium/nickel cations. Further, in-situ X-ray absorption spectroscopic analyses manifest contraction of metal-oxygen bond lengths in the activated catalyst, with a short vanadium-oxygen bond distance. Density functional theory calculations indicate that the vanadium site of the iron/vanadium co-doped nickel (oxy)hydroxide gives near-optimal binding energies of oxygen evolution reaction intermediates and has lower overpotential compared with nickel and iron sites. These findings suggest that the doped vanadium with distorted geometric and disturbed electronic structures makes crucial contribution to high activity of the trimetallic catalyst.

A Survey of Channel Modeling for UAV Communications

Abstract: Unmanned aerial vehicles (UAVs) have attracted great interest in rapid deployment for both civil and military applications. UAV communication has its own distinctive channel characteristics compared to the widely used cellular or satellite systems. Accurate channel characterization is crucial for the performance optimization and design of efficient UAV communication. However, several challenges exist in UAV channel modeling. For example, the propagation characteristics of UAV channels are under explored for spatial and temporal variations in non–stationary channels. Additionally, airframe shadowing has not yet been investigated for small size rotary UAVs. This paper provides an extensive survey of the measurement methods proposed for UAV channel modeling that use low altitude platforms and discusses various channel characterization efforts. We also review from a contemporary perspective of UAV channel modeling approaches, and outline future research challenges in this domain.

Current and future trends in topology optimization for additive manufacturing

Abstract: Manufacturing-oriented topology optimization has been extensively studied the past two decades, in particular for the conventional manufacturing methods, for example, machining and injection molding or casting. Both design and manufacturing engineers have benefited from these efforts because of the close-to-optimal and friendly-to-manufacture design solutions. Recently, additive manufacturing (AM) has received significant attention from both academia and industry. AM is characterized by producing geometrically complex components layer-by-layer, and greatly reduces the geometric complexity restrictions imposed on topology optimization by conventional manufacturing. In other words, AM can make near-full use of the freeform structural evolution of topology optimization. Even so, new rules and restrictions emerge due to the diverse and intricate AM processes, which should be carefully addressed when developing the AM-specific topology optimization algorithms. Therefore, the motivation of this perspective paper is to summarize the state-of-art topology optimization methods for a variety of AM topics. At the same time, this paper also expresses the authors' perspectives on the challenges and opportunities in these topics. The hope is to inspire both researchers and engineers to meet these challenges with innovative solutions.

Facile Ammonia Synthesis from Electrocatalytic N2 Reduction under Ambient Conditions on N-Doped Porous Carbon

Abstract: Ammonia has been used in important areas such as agriculture and clean energy. Its synthesis from the electrochemical reduction of N2 is an attractive alternative to the industrial method that requires high temperature and pressure. Currently, electrochemical N2 fixation has suffered from slow kinetics due to the difficulty of N2 adsorption and N≡N cleavage. Here, N-doped porous carbon (NPC) is reported as a cost-effective electrocatalyst for ammonia synthesis from electrocatalytic N2 reduction under ambient conditions, where its N content and species were tuned to enhance N2 chemical adsorption and N≡N cleavage. The resulting NPC was effective for fixing N2 to ammonia with a high ammonia production rate (1.40 mmol g–1 h–1 at −0.9 V vs RHE). Experiments combined with density functional theory calculations revealed pyridinic and pyrrolic N were active sites for ammonia synthesis and their contents were crucial for promoting ammonia production on NPC. The energy-favorable pathway for ammonia synthesis was *...

Integrated Networking, Caching, and Computing for Connected Vehicles: A Deep Reinforcement Learning Approach

Abstract: The developments of connected vehicles are heavily influenced by information and communications technologies, which have fueled a plethora of innovations in various areas, including networking, caching, and computing. Nevertheless, these important enabling technologies have traditionally been studied separately in the existing works on vehicular networks. In this paper, we propose an integrated framework that can enable dynamic orchestration of networking, caching, and computing resources to improve the performance of next generation vehicular networks. We formulate the resource allocation strategy in this framework as a joint optimization problem, where the gains of not only networking but also caching and computing are taken into consideration in the proposed framework. The complexity of the system is very high when we jointly consider these three technologies. Therefore, we propose a novel deep reinforcement learning approach in this paper. Simulation results with different system parameters are presented to show the effectiveness of the proposed scheme.

Energy-Latency Tradeoff for Energy-Aware Offloading in Mobile Edge Computing Networks

Abstract: Mobile edge computing (MEC) brings computation capacity to the edge of mobile networks in close proximity to smart mobile devices (SMDs) and contributes to energy saving compared with local computing, but resulting in increased network load and transmission latency. To investigate the tradeoff between energy consumption and latency, we present an energy-aware offloading scheme, which jointly optimizes communication and computation resource allocation under the limited energy and sensitive latency. In this paper, single and multicell MEC network scenarios are considered at the same time. The residual energy of smart devices’ battery is introduced into the definition of the weighting factor of energy consumption and latency. In terms of the mixed integer nonlinear problem for computation offloading and resource allocation, we propose an iterative search algorithm combining interior penalty function with D.C. (the difference of two convex functions/sets) programming to find the optimal solution. Numerical results show that the proposed algorithm can obtain lower total cost (i.e., the weighted sum of energy consumption and execution latency) comparing with the baseline algorithms, and the energy-aware weighting factor is of great significance to maintain the lifetime of SMDs.

Motor Anomaly Detection for Unmanned Aerial Vehicles Using Reinforcement Learning

Abstract: Unmanned aerial vehicles (UAVs) are used in many fields including weather observation, farming, infrastructure inspection, and monitoring of disaster areas. However, the currently available UAVs are prone to crashing. The goal of this paper is the development of an anomaly detection system to prevent the motor of the drone from operating at abnormal temperatures. In this anomaly detection system, the temperature of the motor is recorded using DS18B20 sensors. Then, using reinforcement learning, the motor is judged to be operating abnormally by a Raspberry Pi processing unit. A specially built user interface allows the activity of the Raspberry Pi to be tracked on a Tablet for observation purposes. The proposed system provides the ability to land a drone when the motor temperature exceeds an automatically generated threshold. The experimental results confirm that the proposed system can safely control the drone using information obtained from temperature sensors attached to the motor.

Z-Scheme Photocatalytic Water Splitting on a 2D Heterostructure of Black Phosphorus/Bismuth Vanadate Using Visible Light.

Abstract: Spontaneously solar-driven water splitting to produce H2 and O2 , that is, the conversion of solar energy to chemical energy is a dream of mankind. However, it is difficult to make overall water splitting feasible without using any sacrificial agents and external bias. Drawing inspiration from nature, a new artificial Z-scheme photocatalytic system has been designed herein based on the two-dimensional (2D) heterostructure of black phosphorus (BP)/bismuth vanadate (BiVO4 ). An effective charge separation makes possible the reduction and oxidation of water on BP and BiVO4 , respectively. The optimum H2 and O2 production rates on BP/BiVO4 were approximately 160 and 102 μmol g-1 h-1 under irradiation of light with a wavelength longer than 420 nm, without using any sacrificial agents or external bias.

A short review of recent advances in CO2 hydrogenation to hydrocarbons over heterogeneous catalysts

Abstract: CO2 hydrogenation to hydrocarbons is a promising way of making waste to wealth and energy storage, which also solves the environmental and energy issues caused by CO2 emissions Much efforts and research are aimed at the conversion of CO2 via hydrogenation to various value-added hydrocarbons, such as CH4, lower olefins, gasoline, or long-chain hydrocarbons catalyzed by different catalysts with various mechanisms This review provides an overview of advances in CO2 hydrogenation to hydrocarbons that have been achieved recently in terms of catalyst design, catalytic performance and reaction mechanism from both experiments and density functional theory calculations In addition, the factors influencing the performance of catalysts and the first C–C coupling mechanism through different routes are also revealed The fundamental factor for product selectivity is the surface H/C ratio adjusted by active metals, supports and promoters Furthermore, the technical and application challenges of CO2 conversion into useful fuels/chemicals are also summarized To meet these challenges, future research directions are proposed in this review

A Bi-Directional Message Passing Model for Salient Object Detection

Abstract: Recent progress on salient object detection is beneficial from Fully Convolutional Neural Network (FCN). The saliency cues contained in multi-level convolutional features are complementary for detecting salient objects. How to integrate multi-level features becomes an open problem in saliency detection. In this paper, we propose a novel bi-directional message passing model to integrate multi-level features for salient object detection. At first, we adopt a Multi-scale Context-aware Feature Extraction Module (MCFEM) for multi-level feature maps to capture rich context information. Then a bi-directional structure is designed to pass messages between multi-level features, and a gate function is exploited to control the message passing rate. We use the features after message passing, which simultaneously encode semantic information and spatial details, to predict saliency maps. Finally, the predicted results are efficiently combined to generate the final saliency map. Quantitative and qualitative experiments on five benchmark datasets demonstrate that our proposed model performs favorably against the state-of-the-art methods under different evaluation metrics.

Detect Globally, Refine Locally: A Novel Approach to Saliency Detection

Abstract: Effective integration of contextual information is crucial for salient object detection. To achieve this, most existing methods based on 'skip' architecture mainly focus on how to integrate hierarchical features of Convolutional Neural Networks (CNNs). They simply apply concatenation or element-wise operation to incorporate high-level semantic cues and low-level detailed information. However, this can degrade the quality of predictions because cluttered and noisy information can also be passed through. To address this problem, we proposes a global Recurrent Localization Network (RLN) which exploits contextual information by the weighted response map in order to localize salient objects more accurately. Particularly, a recurrent module is employed to progressively refine the inner structure of the CNN over multiple time steps. Moreover, to effectively recover object boundaries, we propose a local Boundary Refinement Network (BRN) to adaptively learn the local contextual information for each spatial position. The learned propagation coefficients can be used to optimally capture relations between each pixel and its neighbors. Experiments on five challenging datasets show that our approach performs favorably against all existing methods in terms of the popular evaluation metrics.

MXene: An emerging material for sensing and biosensing

Abstract: MXene has emerged as an amazing family of two dimensional (2D) layered materials and drawn great attention from researchers of diverse scientific fields. MXenes are the recent advancements of materials chemistry which include early transition metal carbides, nitrides and carbonitrides produced by exfoliation of selective MAX phases. MAX phase corresponds to the general formula Mn+1AXn (n = 1, 2, 3) where M represents early d-block transition metals, A stands for main group sp elements (specifically groups 13 and 14) and X is either C or N atoms. MXenes have left a prodigious impact on scientific communities with new technological advancements for a plethora of potential applications in the field of catalysis, clean energy, electronics, fuel cells, supercapacitors etc. With high metallic conductivity, hydrophilicity, low diffusion barrier, high ion transport properties, biocompatibility, large surface area and ease of functionalization, the MXenes act as fascinating interface for designing next generation detection systems exploiting their utilization in analytical chemistry. Recent progress in the field of MXenes emphasizing their significant role in analytical sensing has been well discussed in this review. Future perspectives with a motivated research in the field of MXenes based sensors have been focused at the end. The underlying goal of this review is to acquaint the readers with the sensing applications of MXenes and to document the latest advancements made in this area till date.

High-Density Ultra-small Clusters and Single-Atom Fe Sites Embedded in Graphitic Carbon Nitride (g-C3N4) for Highly Efficient Catalytic Advanced Oxidation Processes

Abstract: Ultra-small metal clusters have attracted great attention owing to their superior catalytic performance and extensive application in heterogeneous catalysis However, the synthesis of high-density metal clusters is very challenging due to their facile aggregation Herein, one-step pyrolysis was used to synthesize ultra-small clusters and single-atom Fe sites embedded in graphitic carbon nitride with high density (iron loading up to 182 wt %), evidenced by high-angle annular dark field-scanning transmission electron microscopy, X-ray absorption spectroscopy, X-ray photoelectron spectroscopy, and 57Fe Mossbauer spectroscopy The catalysts exhibit enhanced activity and stability in degrading various organic samples in advanced oxidation processes The drastically increased metal site density and stability provide useful insights into the design and synthesis of cluster catalysts for practical application in catalytic oxidation reactions

Aggregation-Resistant 3D MXene-Based Architecture as Efficient Bifunctional Electrocatalyst for Overall Water Splitting.

Abstract: The MXene combining high conductivity, hydrophilic surface, and wide chemical variety has been recognized as a rapidly rising star on the horizon of two-dimensional (2D) material science. However, strong tendency to intersheet aggregate via van der Waals force represents a major problem limiting the functionalities, processability, and performance of MXene-based material/devices. We report a capillary-forced assembling strategy for processing MXene to hierarchical 3D architecture with geometry-based high resistance to aggregation. Aggregate-resistant properties of 3D MXene not only double the surface area without loss of the intrinsic properties of MXene but also render the characteristics such as kinetics-favorable framework, high robustness, and excellent processability in both solution and solid state. Synergistically coupling the 3D MXene with electrochemically active phases such as metal oxide/phosphides, noble metals, or sulfur yields the hybrid systems with greatly boosted active surface area, charge-transfer kinetics, and mass diffusion rate. Specifically, the CoP-3D MXene hybrids exhibit high electrocatalytic activity toward oxygen and hydrogen evolution in alkaline electrolyte. As a bifunctional electrocatalyst, they exhibit superior cell voltage and durability to combined RuO2/Pt catalysts for overall water splitting in basic solution, highlighting the great promise of aggregation-resistant 3D MXene in the development of high-performance electrocatalysts.

Near-Infrared Light-Initiated Molecular Superoxide Radical Generator: Rejuvenating Photodynamic Therapy against Hypoxic Tumors

Abstract: Hypoxia, a quite universal feature in most solid tumors, has been considered as the “Achilles’ heel” of traditional photodynamic therapy (PDT) and substantially impairs the overall therapeutic efficacy. Herein, we develop a near-infrared (NIR) light-triggered molecular superoxide radical (O2–•) generator (ENBS-B) to surmount this intractable issue, also reveal its detailed O2–• action mechanism underlying the antihypoxia effects, and confirm its application for in vivo targeted hypoxic solid tumor ablation. Photomediated radical generation mechanism study shows that, even under severe hypoxic environment (2% O2), ENBS-B can generate considerable O2–• through type I photoreactions, and partial O2–• is transformed to high toxic OH· through SOD-mediated cascade reactions. These radicals synergistically damage the intracellular lysosomes, which subsequently trigger cancer cell apoptosis, presenting a robust hypoxic PDT potency. In vitro coculture model shows that, benefiting from biotin ligand, ENBS-B achieve...

Multiview Spectral Clustering via Structured Low-Rank Matrix Factorization

Abstract: Multiview data clustering attracts more attention than their single-view counterparts due to the fact that leveraging multiple independent and complementary information from multiview feature spaces outperforms the single one. Multiview spectral clustering aims at yielding the data partition agreement over their local manifold structures by seeking eigenvalue–eigenvector decompositions. Among all the methods, low-rank representation (LRR) is effective, by exploring the multiview consensus structures beyond the low rankness to boost the clustering performance. However, as we observed, such classical paradigm still suffers from the following stand-out limitations for multiview spectral clustering of overlooking the flexible local manifold structure, caused by aggressively enforcing the low-rank data correlation agreement among all views, and such a strategy, therefore, cannot achieve the satisfied between-views agreement; worse still, LRR is not intuitively flexible to capture the latent data clustering structures. In this paper, first, we present the structured LRR by factorizing into the latent low-dimensional data-cluster representations, which characterize the data clustering structure for each view. Upon such representation, second, the Laplacian regularizer is imposed to be capable of preserving the flexible local manifold structure for each view. Third, we present an iterative multiview agreement strategy by minimizing the divergence objective among all factorized latent data-cluster representations during each iteration of optimization process, where such latent representation from each view serves to regulate those from other views, and such an intuitive process iteratively coordinates all views to be agreeable. Fourth, we remark that such data-cluster representation can flexibly encode the data clustering structure from any view with an adaptive input cluster number. To this end, finally, a novel nonconvex objective function is proposed via the efficient alternating minimization strategy. The complexity analysis is also presented. The extensive experiments conducted against the real-world multiview data sets demonstrate the superiority over the state of the arts.

Superhierarchical Cobalt-Embedded Nitrogen-Doped Porous Carbon Nanosheets as Two-in-One Hosts for High-Performance Lithium–Sulfur Batteries

Abstract: Lithium-sulfur (Li-S) batteries, based on the redox reaction between elemental sulfur and lithium metal, have attracted great interest because of their inherently high theoretical energy density. However, the severe polysulfide shuttle effect and sluggish reaction kinetics in sulfur cathodes, as well as dendrite growth in lithium-metal anodes are great obstacles for their practical application. Herein, a two-in-one approach with superhierarchical cobalt-embedded nitrogen-doped porous carbon nanosheets (Co/N-PCNSs) as stable hosts for both elemental sulfur and metallic lithium to improve their performance simultaneously is reported. Experimental and theoretical results reveal that stable Co nanoparticles, elaborately encapsulated by N-doped graphitic carbon, can work synergistically with N heteroatoms to reserve the soluble polysulfides and promote the redox reaction kinetics of sulfur cathodes. Moreover, the high-surface-area pore structure and the Co-enhanced lithiophilic N heteroatoms in Co/N-PCNSs can regulate metallic lithium plating and successfully suppress lithium dendrite growth in the anodes. As a result, a full lithium-sulfur cell constructed with Co/N-PCNSs as two-in-one hosts demonstrates excellent capacity retention with stable Coulombic efficiency.

Offloading in Internet of Vehicles: A Fog-Enabled Real-Time Traffic Management System

Abstract: Fog computing has been merged with Internet of Vehicle (IoV) systems to provide computational resources for end users, by which low latency can be guaranteed. In this paper, we put forward a feasible solution that enables offloading for real-time traffic management in fog-based IoV systems, aiming to minimize the average response time for events reported by vehicles. First, we construct a distributed city-wide traffic management system, in which vehicles close to road side units can be utilized as fog nodes. Then, we model parked and moving vehicle-based fog nodes according to a queueing theory, and draw the conclusion that moving vehicle-based fog nodes can be modeled as an $M/M/1$ queue. An approximate approach is developed to solve the offloading optimization problem by decomposing it into two subproblems and scheduling traffic flows among different fog nodes. Performance analyses based on a real-world taxi-trajectory datasets are conducted to illustrate the superiority of our method.

How Can Heterogeneous Internet of Things Build Our Future: A Survey

Abstract: Heterogeneous Internet of Things (HetIoT) is an emerging research field that has strong potential to transform both our understanding of fundamental computer science principles and our future living. HetIoT is being employed in increasing number of areas, such as smart home, smart city, intelligent transportation, environmental monitoring, security systems, and advanced manufacturing. Therefore, relaying on strong application fields, HetIoT will be filled in our life and provide a variety of convenient services for our future. The network architectures of IoT are intrinsically heterogeneous, including wireless sensor network, wireless fidelity network, wireless mesh network, mobile communication network, and vehicular network. In each network unit, smart devices utilize appropriate communication methods to integrate digital information and physical objects, which provide users with new exciting applications and services. However, the complexity of application requirements, the heterogeneity of network architectures and communication technologies impose many challenges in developing robust HetIoT applications. This paper proposes a four-layer HetIoT architecture consisting of sensing, networking, cloud computing, and applications. Then, the state of the art in HetIoT research and applications have been discussed. This paper also suggests several potential solutions to address the challenges facing future HetIoT, including self-organizing, big data transmission, privacy protection, data integration and processing in large-scale HetIoT.

Dendritic core-shell nickel-iron-copper metal/metal oxide electrode for efficient electrocatalytic water oxidation.

Abstract: Electrochemical water splitting requires efficient water oxidation catalysts to accelerate the sluggish kinetics of water oxidation reaction. Here, we report a promisingly dendritic core-shell nick ...

Direct Observation of Structural Evolution of Metal Chalcogenide in Electrocatalytic Water Oxidation

Abstract: As one of the most remarkable oxygen evolution reaction (OER) electrocatalysts, metal chalcogenides have been intensively reported during the past few decades because of their high OER activities. It has been reported that electron-chemical conversion of metal chalcogenides into oxides/hydroxides would take place after the OER. However, the transition mechanism of such unstable structures, as well as the real active sites and catalytic activity during the OER for these electrocatalysts, has not been understood yet; therefore a direct observation for the electrocatalytic water oxidation process, especially at nano or even angstrom scale, is urgently needed. In this research, by employing advanced Cs-corrected transmission electron microscopy (TEM), a step by step oxidational evolution of amorphous electrocatalyst CoS x into crystallized CoOOH in the OER has been in situ captured: irreversible conversion of CoS x to crystallized CoOOH is initiated on the surface of the electrocatalysts with a morphology change via Co(OH)2 intermediate during the OER measurement, where CoOOH is confirmed as the real active species. Besides, this transition process has also been confirmed by multiple applications of X-ray photoelectron spectroscopy (XPS), in situ Fourier-transform infrared spectroscopy (FTIR), and other ex situ technologies. Moreover, on the basis of this discovery, a high-efficiency electrocatalyst of a nitrogen-doped graphene foam (NGF) coated by CoS x has been explored through a thorough structure transformation of CoOOH. We believe this in situ and in-depth observation of structural evolution in the OER measurement can provide insights into the fundamental understanding of the mechanism for the OER catalysts, thus enabling the more rational design of low-cost and high-efficient electrocatalysts for water splitting.

UAV Trajectory Optimization for Data Offloading at the Edge of Multiple Cells

Abstract: In future mobile networks, it is difficult for static base stations (BSs) to support the rapidly increasing data services, especially for cell-edge users. Unmanned aerial vehicle (UAV) is a promising method that can assist BSs to offload the data traffic, due to its high mobility and flexibility. In this correspondence, we focus on the UAV trajectory at the edges of three adjacent cells to offload traffic for BSs. In the proposed scheme, the sum rate of UAV-served edge users is maximized subject to the rate requirements for all the users, by optimizing the UAV trajectory in each flying cycle. The optimization is a mixed-integer nonconvex problem, which is difficult to solve. Thus, it is transformed into two convex problems, and an iterative algorithm is proposed to solve it by optimizing the UAV trajectory and edge user scheduling alternately. Simulation results are presented to show the effectiveness of the proposed scheme.